SEM Price Optimization: Balancing Human Input Costs

It’s difficult to express the value of SEM Optimization because the mathematics / economics behind it is hard to express in a simple Y = mx + b format.

The valuation of SEM Labor Input Optimization does not lend itself to that.

The least controversial ‘law’ is that the cost of human administration increases as the SEM budget increases. For instance, a small budget of 10000 dollars a year wouldn’t require all that much human administration – by and large setting up a single platform. However, as SEM budgets expand, out to 30 grand up through 1 million and beyond, you need more platforms, and ultimately out into affiliate programs – programs that require a lot more human administration.

Human hours scale with budget, and that expansion can be fairly lumpy. In principle though, pure administration budgets range from 6.5% to 10% of a SEM budget. The more you spend online, the more you’ll pay in terms of labor administration cost. There’s not free lunch here.

The second fundamental law is that the greater your budget, the higher your Cost Per Conversion is going inflate.

This one is a little bit harder to swallow.

It’s possible, at very low budget numbers, to achieve very efficient CPC’s, because the amount of micro-inventory is quite specialized. If you’re running a campaign for gilded pens – with very low budgets, you can target very small communities and keywords, and really minimize that CPC. They’re interested. Clickthrough is high and conversion is very good.

At some point though, you run out of specialized inventory. You have to go out and start spending more, in looser interested communities and keywords. Your diluting the effectiveness of your spend because you’re increasing your reach. This relationship isn’t necessarily linear either. If your niche target audience is 10000 people, you’re going to start diluting your effectiveness a lot sooner than if your niche target audience is 500000.

The third relationship is easier.

The more time a human spends planning and optimizing a campaign, the more efficient it is. The lower the CPC. For instance, a lovely architected plan will identify and target very specific keywords. The effort that goes into constructing a keyword matrix, or sifting through clickstream and performance data for where things are falling apart, is not to be trivialized. The nature of those efforts are ‘lumpy’. Sometimes you find gold on your first pan. Sometimes you could be out at the river all day and find nothing more than a few specs of gold dust. It’s a process that’s worth it.

Let’s also accept that there’s some natural floor, and that after a certain point, a human could spend more and more time, and hit that limit. In fact, such an absolute floor does exist – it would be impossible to buy ad space with a cost of zero and a conversion of 100% – even when you take into account social or viral. There’s a definitive cost to that.

So, there exists a marginal effort curve. Hiring an expert for the first 10 hours will yield really strong results. The next 30 hours may yield the same equivalent result of the first 10. The next 300 hours equivalent to the gains made in the first 10, and so on. These manifest themselves in a reduction in cost per acquisition.

Consider:

As you spend more money on humans to optimize your spend, your cost per conversion falls.

As you spend more money in your overall budget, the more money you need to spend on humans to administer it.

As you spend more money in your overall budget, total, your cost per conversion increases.

I loaded these into a grapher, and made assumptions around CPC, the marginal effectiveness of optimization, the marginal cost of administration, and the dilution rate of budget.

What would normally follow is an expansive set of data – but I think, that I’ll throw these 3 ‘laws’ or ‘postulates’ out there, and see who takes up the fight.